This invention relates to a resist pattern-forming apparatus which bakes a substrate having a developed resist pattern and descums said pattern.
A substrate on which the predetermined resist pattern is formed after light projection and development, and which is used as an etching mask, undergoes a baking process to eliminate the developing solution deposited on the resist pattern formed on the substrate, and also undergoes a descumming process to remove the scum which has settled on said resist pattern, and is later used as an etching mask for a semiconductor substrate. In the baking step, a substrate on which a resist pattern is formed is put into an oven kept at a temperature of, for example, 100.degree..+-.5.degree. C., for 10 to 60 minutes. Later in the descumming step, the substrate which has been fully baked is allowed to cool in the air for 60 minutes. The substrate undergoes plasma treatment for 1 to 10 minutes in an atmosphere of moist oxygen gas which is held in a vacuum chamber, has a pressure of pb 0.1 to 0.9 torr, and is held at a temperature of 25.degree..+-.10.degree. C. As a result, said substrate is descummed.
However, the conventional process has the drawbacks, in that the baking of the substrate and the cooling of the baked substrate consumes 1 to 2 hours, and moreover, an operator has to carry the baked substrate from the oven to the descumming apparatus. Consequently, the operation of baking and later descumming the substrate is carried out very inefficiently. When a certain number of substrates are descummed one after another, the temperature in the vacuum chamber is raised from the room temperature to about 40.degree. C. by the action of plasma. As a result, the substrate temperature also rises, resulting in a noticeable increase in the descumming rate of the resist pattern. This event causes the resist pattern dimensions to vary up to a maximum of about 0.5 microns before and after the pattern is descummed. A method has been proposed to operate the descumming apparatus intermittently in order to avoid the above-mentioned variations in the measurements of the resist pattern. However, this proposed method has the drawback that a waiting time of between 10 and 60 minutes has to elapse between the time when one substrate is descummed and the time at which the succeeding substrate is supplied for the descumming operation-a tremendous decline in descumming efficiency.
The above-mentioned drawbacks are observable also in the case where ozone is taken into the vacuum chamber from an ozone generator, to carry out the descumming operation.